Enterprise AI Scaling Encounters Difficulties in Research Reports
In a recent report, Gartner, a leading research and advisory company, has highlighted the challenges facing enterprise AI, particularly in the realm of agentic AI. According to the report, titled "Emerging Tech: Avoid Agentic AI Failure: Build Success Using Right Use Cases," 95% of organizations are currently getting zero return on their GenAI investment.
Anushree Verma, Senior Director Analyst at Gartner, emphasizes that most agentic AI projects are early-stage experiments or proof of concepts, often driven by hype and misapplied. Sixty percent of organizations have evaluated enterprise-grade AI systems, but only 20% have reached the pilot stage, and just 5% have made it to production.
The report underscores the importance of organizing structured and unstructured data into a well-modeled repository, such as a knowledge graph, before layering in AI. Providing precise, business-aligned instructions for AI agents is also crucial to ensure consistent, business-focused performance.
The foundation of successful AI agent deployment is data quality, with the adage "Garbage in, garbage out" being particularly relevant. Prioritizing trust through model/data pairing, using well-selected language models with high-fidelity operational data, supports reliable decision-making.
The report further notes that AI agents are well-suited for complex, variable-rich industrial environments, aiding in parsing documentation, identifying pertinent information, and integrating it into data models. However, most failures of enterprise-grade AI systems are due to brittle workflows, lack of contextual learning, and misalignment with day-to-day operations.
The report also highlights the three pillars essential to trustworthy agent operations: selecting the right language model, ensuring agents access only vetted, context-rich data, and crafting specific, clear instructions. Building an industrial knowledge graph offers immediate benefits, including enhanced dashboarding, faster query response times, better troubleshooting, and overall productivity gains across operations.
Interestingly, over 80% of organizations have explored or piloted tools like ChatGPT and Copilot, but these primarily enhance individual productivity, not P&L performance. This finding suggests a significant challenge in scaling enterprise AI, with a focus on enterprise productivity being necessary rather than just individual task augmentation.
In a positive note, only 5% of integrated AI pilots are extracting millions in value, according to the MIT study, "The GenAI Divide: State of AI in Business 2025." The study, based on a review of over 300 publicly disclosed AI initiatives, interviews with representatives from 52 organizations, and survey responses from 153 senior leaders collected across four major industry conferences, underscores the potential for successful AI implementation in the enterprise.
However, the road to success is not without its hurdles. The Gartner report and the MIT study both underscore the need for careful planning, data modelling, and clear, business-focused instructions to ensure the successful deployment of AI agents in the enterprise.
Please note that the Gartner report is only available to Gartner clients. For more insights, it is recommended that interested parties consider becoming Gartner clients or reaching out to the company for further information.
